Down­load the full inter­view here. 

Robert Stephens has spent 35 years in oil and gas, where a sys­tem glitch can do more than dent the bot­tom line; it can put peo­ple at risk. For the past decade, he has been a Prin­ci­pal at PwC U.S., guid­ing SAP pro­grams for the company’s clients with those stakes in mind.

At Cono­coPhillips, Stephens led the tran­si­tion from SAP ECC to SAP S/4HANA with a clean-core man­date: strip out mil­lions of lines of cus­tom code, keep process­es fit to stan­dard, and shift the few nec­es­sary exten­sions onto SAP BTP. The effort also con­sol­i­dat­ed 25 sep­a­rate lega­cy data ware­hous­es into a sin­gle glob­al cloud-based instance on Microsoft Azure and expand­ed the use of dig­i­tal twins to reduce time spent on high-risk assets.

The result, Stephens argues, is proof for oth­er SAP cus­tomers that clean core is not a lim­i­ta­tion but a foun­da­tion that can low­er cost to serve, steady oper­a­tions, and cre­ate room for growth. Below, Stephens dis­cuss­es the under­tak­ing at Cono­coPhillips, pro­vid­ing an inside look at the transformation.

This inter­view has been edit­ed and con­densed for length and clarity.

Q: With regard to the wider pic­ture at Cono­coPhillips, what was dri­ving their deci­sion to embark on a busi­ness trans­for­ma­tion in the first place?

So that was twofold, one part busi­ness-dri­ven and the sec­ond part tech­nol­o­gy-dri­ven. On the busi­ness side, their cost to serve and cost to pro­duce were grow­ing at a faster rate than pro­duc­tion was.

Let me rephrase that: they want­ed to look at ways to dri­ve SG&A cost out of the equa­tion. A big part of their SG&A cost was IT cost. That led to what they need­ed to do from an IT per­spec­tive to be a low­er cost to serve their inter­nal con­stituents, serve the com­pa­ny, and ulti­mate­ly help serve their shareholders.

From anoth­er out­side influ­ence, SAP announced the move to SAP S/4HANA. When this project was in its infan­cy, before it had a name or a clear direc­tion, the first tar­get was 2020, the first time SAP announced that all cus­tomers had to be off SAP ECC by 2020. As we got to the 2016 to 2017 time­frame, SAP moved the end date for stan­dard sup­port for SAP ECC out to 2025.

Then it became 2027. This ongo­ing saga of SAP mov­ing the SAP ECC end date drove the tim­ing of the project. When they decid­ed to fund the project in 2021, it was about get­ting to a low­er cost to serve. One of the key dri­vers was get­ting back to fit-to-stan­dard across their appli­ca­tion port­fo­lio and adopt­ing cloud tech­nolo­gies. They felt those two things were fun­da­men­tal to estab­lish­ing a tech­nol­o­gy foun­da­tion for growth at a rea­son­able rate of investment.

Request for Pro­pos­als (RFPs) to the soft­ware ven­dors and to the Sys­tem Inte­gra­tor (SI) ven­dors went out in 2020. Selec­tions were made, and the project start­ed in Sep­tem­ber 2021.

Q: In our pre­vi­ous con­ver­sa­tion, you talked about the prin­ci­ples of on time, on bud­get, and no oper­a­tional impact. How did those play out in prac­tice and guide your approach?

The time­line often dri­ves the bud­get. You have scope and you have a time­line, and those are the key fac­tors. The scope was rel­a­tive­ly con­stant across the pro­gram, but the time­line had one major adjust­ment. Dur­ing the com­pet­i­tive phas­es, we laid out a pro­gram we thought was right for Cono­coPhillips at some­where between 36 and 39 months.

As we were being con­sid­ered down and select­ed, Cono­coPhillips chal­lenged us to plan around a 30-month imple­men­ta­tion. We start­ed glob­al design, moved through glob­al design, and start­ed our glob­al build. As we came up on the first go-live, which was Cana­da, Cono­coPhillips real­ized after the first 18 months that they could not con­sume all the change we had intro­duced from fit-to-stan­dard, cloud adop­tion, and orga­ni­za­tion­al change.

They asked us to extend the time­line for roll­out. We went back, did a rebase­line, and set­tled on a 39-month pro­gram. We added a year to the roll­out sched­ule. Once we rebase­lined, we met each deliv­ery mile­stone, each go-live, and the bud­get asso­ci­at­ed with the rebase­line. Although the time­line and bud­get changed, it was an agreed effort between us as the SI and them as the cus­tomer. They could not con­sume as fast as we could deliver.

Some key data points: Cono­coPhillips nev­er stopped pump­ing oil for a sin­gle day due to the SAP project. They paid their ven­dors, their employ­ees, and their share­hold­ers and roy­al­ty own­ers. There was no impact on operations.

We fin­ished the core pro­gram. The last go-live for the major pro­gram was Jan. 1 of this year, which was U.S. oper­a­tions. At the same time, we went live with U.S. oper­a­tions, and they acquired Marathon Oil. We did the U.S. go-live and start­ed the Marathon inte­gra­tion in parallel.

At the same time, they were doing a shut­down and turn­around of their North Sea oper­a­tions. We had to care­ful­ly plan that. Half of North Sea pro­duc­tion was shut in over a 25-day peri­od, and we could not go a day longer. We orches­trat­ed that from the end of May to the end of June. We came out of that shut­down, did a tech­ni­cal upgrade of the SAP envi­ron­ment on June 28, and went live with the Marathon inte­gra­tion on July 1. That orches­tra­tion, last­ing a 45-day peri­od, was filled with sleep­less nights, but we even­tu­al­ly deliv­ered it.

Ryan Lance, the CEO of Cono­coPhillips, did the quar­ter­ly earn­ings call and said, ​“You said we could not inte­grate Marathon in sev­en months, and you said we could not deliv­er a bil­lion dol­lars in effi­cien­cies, and we just did it at the same time because of our tech­nol­o­gy plat­form being in place that allowed us to oper­ate flawlessly.”

Q: What are some of the chal­lenges of oil and gas imple­men­ta­tions com­pared to oth­er indus­tries you have worked with? 

I am com­ing up on my 35-year career anniver­sary, and I have spent almost my entire career serv­ing the oil and gas industry.

In oil and gas, if we make mis­takes, peo­ple can get hurt. In con­sumer goods, if a mis­take is made, a store shelf might go emp­ty, but it is unlike­ly that any­one will get hurt because of that. In oil and gas, util­i­ties, trav­el, and trans­porta­tion, if you make a major mis­take, peo­ple get hurt. That is what you want to avoid. I don’t think we are spe­cial. We just have to be more care­ful than some oth­er industries.

Q: You said dif­fer­en­ti­a­tion is not a free pass for over cus­tomiz­ing. How did you com­mu­ni­cate the need to fit the stan­dard and avoid unnec­es­sary cus­tomiza­tions, and how did you apply that through­out this project?

Cono­coPhillips, in one form or anoth­er, has been on SAP since 1998. In the SAP R/3 and SAP ECC world, if the busi­ness asked for it, they built it in SAP. They end­ed with some­where between eight and nine mil­lion lines of cus­tom code. This was a philo­soph­i­cal and cul­tur­al change to go back to fit-to-stan­dard and clean core.

The envi­ron­ment they were com­ing from was incred­i­bly expen­sive to oper­ate, main­tain, and upgrade. The only way to dri­ve cost out was to dri­ve cus­tomiza­tion out. To get to a low­er cost to serve, they had to make that change. In our capa­bil­i­ty-dri­ven strat­e­gy, we char­ac­ter­ized each busi­ness process as dif­fer­en­ti­at­ing, com­pet­i­tive­ly nec­es­sary, or foun­da­tion­al. If it were foun­da­tion­al, there was no ques­tion that there would be no cus­tomiza­tion. We would get back to fit-to-stan­dard. As we moved up to com­pet­i­tive­ly nec­es­sary and dif­fer­en­ti­at­ing, the con­ver­sa­tion got deep­er, but dif­fer­en­ti­at­ing did not mean customization.

We chal­lenged them to adopt the mind­set that dif­fer­en­ti­at­ing does not mean cus­tomiza­tion. We deliv­ered the core pro­gram at around 94% stan­dard out of the box. With the most recent upgrade to the 2023 Fea­ture Pack 03, Cono­coPhillips is now at 97% stan­dard out of the box.

I would argue that it is top tier in any industry.

Q: Why was SAP BTP part of the core project as a foun­da­tion­al element?

When we did need exten­si­bil­i­ty, we moved the remain­ing 3% to SAP BTP to keep the core clean. We took advan­tage of the plat­form and moved exten­si­bil­i­ty out of the core. We did have to build some BAPIs and sim­i­lar items in the core where things were not yet avail­able. Even in those ele­ments where we had to cus­tomize, we fol­lowed the SAP clean-core mind­set. I believe we have deliv­ered a solu­tion that close­ly aligns with the high stan­dards expect­ed for a clean core in the indus­try and pri­vate cloud.

Q: Could you talk about hav­ing clar­i­ty on SAP’s direc­tion and com­mu­ni­cat­ing that to stake­hold­ers, espe­cial­ly for indus­tries that are not always sure how to stay on the roadmap?

We embraced SAP as part of our core project team. They did not have a huge num­ber of peo­ple on the team, but they had mem­bers in key areas where we were press­ing the limit.

A new prod­uct released in 2021, field logis­tics, replaced the old remote logis­tics man­age­ment solu­tion. It was a com­plete rewrite by SAP. We had the prod­uct team embed­ded in our deliv­ery. Rather than devel­op­ing the prod­uct in a black box, they sat in the facil­i­ty with us, and we showed them how we need­ed to use the solu­tion. We ben­e­fit­ed from the solu­tion work­ing the way we need­ed it to, with SAP doing prod­uct devel­op­ment at the back end. That was fundamental.

We also had Peter Maier and Juer­gen Eise­le on our exec­u­tive steer­ing com­mit­tee. Each time we reviewed a devi­a­tion from stan­dard func­tion­al­i­ty, the first ques­tion to SAP was whether it was on the roadmap. If yes, when? Then we could decide whether to put a workaround in place, do cus­tom devel­op­ment, or do a hybrid while wait­ing for stan­dard functionality.

Con­nect with Robert Stephens on LinkedIn.

If it was not on the roadmap, we used the Cus­tomer Influ­ence por­tal to get items added. In some cas­es, SAP said absolute­ly not, and in those areas, we had to solve it, default­ing to SAP BTP where pos­si­ble. The exec­u­tive steer­ing com­mit­tee had to sign off on each one of them. In a few cas­es, the CFO said we were not that spe­cial in func­tions like trea­sury, and that we should change our process to the stan­dard process and for­get customization.

We had a four-lay­er review from the solu­tion team, pro­gram lead­er­ship, our deci­sion board, and the exec­u­tive steer­ing com­mit­tee. With that many smart peo­ple look­ing at solu­tions, we came up with good ideas.

Q: How did you approach the roll­out sequence, and why did you choose Cana­da first, then Asia Pacif­ic, then Europe and North Amer­i­ca along­side the Marathon acquisition?

We did glob­al design for about nine months, then a glob­al build for about a year, then region­al rollouts.

Cana­da and the U.S. were in the same time zone. The U.S. is the largest busi­ness unit. With a brand-new solu­tion, you don’t take the largest unit first. We picked the small­er of the two in our local time zone to help sta­bi­lize the sys­tem. That is why Cana­da went first.

Next, we looked at the next small­er unit or col­lec­tion of units. Asia Pacif­ic had one oper­at­ed set of assets in Aus­tralia, with every­thing else as non-oper­at­ed joint ven­tures. That made it eas­i­er to con­sume. Cana­da went live in August 2023. In Jan­u­ary 2024, we went live with Asia Pacif­ic. That left two big units. Nor­way had a turn­around com­ing up in 2025, so they need­ed to go in 2024 to have a full year on the new sys­tem before the turn­around. Nor­way and the U.K. went live on July 1, 2024.

That left the Low­er 48 and Alas­ka, which are sep­a­rate busi­ness units, locat­ed in the U.S. We took them live on Jan. 1, 2025. Then the Marathon acqui­si­tion was announced. Marathon’s asset base was a per­fect over­lay into the Low­er 48, which had just gone live in Jan­u­ary, plus assets in Equa­to­r­i­al Guinea that were very sim­i­lar to Norway’s assets. We decid­ed on a large-scale inte­gra­tion in July of this year.

Q: What can you say about the foun­da­tion for data ana­lyt­ics, machine learn­ing, and AI that you built to keep Cono­coPhillips com­pet­i­tive for future AI implementations?

The foun­da­tion for AI and machine learn­ing is good data. Cono­coPhillips, like most orga­ni­za­tions, had a lot of data and a lot of dupli­cates. Their pre­vi­ous strat­e­gy had around 25 on-premis­es lega­cy data ware­hous­es across the globe. As part of this pro­gram, we col­lapsed those mul­ti­ple lega­cy ware­hous­es into a sin­gle glob­al cloud instance on Azure. We cre­at­ed a glob­al data ware­house, one source of truth, and used that as the report­ing base, plus all the oper­a­tional data going in there.

We also imple­ment­ed SAP APM for asset per­for­mance man­age­ment. His­tor­i­cal­ly, they were a planned, sched­uled, and reac­tive main­te­nance orga­ni­za­tion. We have laid the foun­da­tion for pre­dic­tive maintenance.

We did not do a lot of the buzz­wor­thy AI things that many peo­ple are talk­ing about now, because that trend took off mid­way through the pro­gram. We decid­ed not to dis­rupt the pro­gram to change for AI. We stayed true to our prin­ci­ples and deliv­ered the pro­gram. Now they have a plat­form to take advan­tage of AI and machine learning.

They have also been big users of dig­i­tal twins for large assets. Some of that was already in place before the pro­gram, and we enhanced it and con­firmed they did not lose any dig­i­tal twin capa­bil­i­ty across the asset base as we imple­ment­ed the ERP program.

Q: Why was the dig­i­tal twin approach so important?

It goes back to putting peo­ple on the asset.

Peo­ple inter­fac­ing with the asset are the riski­est part of our busi­ness. Putting a per­son on a plat­form, a drill pad, or an oper­at­ing pad is when that per­son is most exposed. If they have a dig­i­tal twin, some­thing they can use in the safe­ty of an office to make oper­a­tional deci­sions, main­te­nance deci­sions, plan­ning, and fore­cast­ing from a dig­i­tal mod­el, we have giv­en them the abil­i­ty to oper­ate more effec­tive­ly and safe­ly. That is why they adopt­ed a cul­ture of dig­i­tal twins.

Q: What are the biggest take­aways, lessons learned, and results you can hang your hat on for Cono­coPhillips and for PwC?

This was one of the largest pro­grams PwC had done since we returned to the SI busi­ness. The ben­e­fit was that SAP S/4HANA was on the hori­zon. We built our SAP busi­ness for the cloud, for SAP S/4HANA, for the future. We were not sup­port­ing a lega­cy SAP ECC business.

Cono­coPhillips was the first large-scale imple­men­ta­tion we had done since restart­ing our prac­tice. Oil and gas can be very sat­u­rat­ed with SAP. SAP used to pub­lish a sta­tis­tic that more than 90% of the world’s oil pro­duc­tion was account­ed for in SAP. I am not sure the num­ber is that large now, but it is high. These com­pa­nies have a long his­to­ry with SAP.

Vis­it the PwC web­site.

We showed that you can do SAP dif­fer­ent­ly in an indus­try sat­u­rat­ed with SAP for 20-plus years. The sto­ry can be com­pelling in the indus­try and across indus­tries. For the fore­see­able future, they are in a cat­e­go­ry of one. They were first to do this, and they have a first-mover advan­tage. As Ryan Lance said, they did a $23.5 bil­lion acqui­si­tion and inte­grat­ed it in six months because of the invest­ment they made in their tech­nol­o­gy platform.

It can give oth­ers con­fi­dence. I talk to a lot of cus­tomers. If Cono­coPhillips can do it with PwC, why can’t I? Cus­tomers are think­ing dif­fer­ent­ly. In SAP, each cus­tomer thinks they are spe­cial. That is why they cus­tomized their envi­ron­ments. Now they are under­stand­ing that you can still be spe­cial and dif­fer­en­ti­at­ed and do great things with a stan­dard prod­uct. I was at a large chem­i­cal com­pa­ny recent­ly, com­pet­ing with anoth­er inte­gra­tor. The oth­er inte­gra­tor told them that a world-class fit-to-stan­dard would be 85%.

I said that is the old way of think­ing. World-class fit-to- stan­dard is 95% plus. I can give you a cus­tomer to talk to who is at plus-97%. I said, ​“Don’t trust me; let my cus­tomer talk to you.” That has ele­vat­ed us to the oppor­tu­ni­ty to win this glob­al imple­men­ta­tion. We are chang­ing the mind­set of what good looks like.